SCEPTICS and believers alike have been struggling for more than 100 years to show whether or not psychic powers exist. Yet still there is nothing even approaching a definitive answer. You might have imagined that a century of work would have led somewhere, so why don鈥檛 we know?
One obvious conclusion is that there is simply nothing to find. Yet that does a disservice to the parapsychologists and their results. Over the past 20 years, they have made real efforts to meet the sceptics鈥� criticisms, designing carefully controlled experiments, carried out under conditions acceptable to scientists (see 鈥淲hat is psi?鈥�), in a genuine attempt to resolve the issue. And many of their experiments have produced results that, on the surface at least, look impressive.
Yet still there is no progress. Yes, there are positive results 鈥� but there are negative ones too. Overall, it鈥檚 hard not to conclude that this mixed bag of data has had little effect other than to fan the flames of debate.
Advertisement
One of the thorniest issues is that of replicability. The ability to repeat an experiment would seem to be a reasonable thing to demand of a field that aspires to scientific respectability. But even replicability has proved a slippery and controversial concept.
Many sceptics argue that only 鈥渞eplication on demand鈥� can provide conclusive proof of the paranormal. Parapsychologists, however, have successfully argued that 鈥渟tatistical replication鈥� is good enough 鈥� that the effect should be replicable more often than you would expect by chance, as revealed by a 鈥渕eta-analysis鈥� of all the available data.
In practice, however, meta-analysis has failed to produce a consensus, largely because its methods contain so much room for manoeuvre. Take a widely used experimental technique in ESP research called ganzfeld, for example. Between 1995 and 1999, approximately 30 ganzfeld studies were carried out. The results from these studies have been subject to four different meta-analyses. Two conclude that the findings are significant; two that they are not.
Why the discrepancy? There are many minor differences between the four analyses, but one stands out above others: whether or not to include a hugely successful study carried out in 1997 by Kathy Dalton of the University of Edinburgh in the UK. Two meta-analyses left it out on the grounds that its results were so much better than any others that the study should be discounted as an outlier 鈥� an acceptable practice in meta-analyses. The other two included it on the grounds that meta-analyses must use all the data 鈥� again, an acceptable practice. No prizes for guessing which analyses came to which conclusions.
Another typical episode concerns the work of Robert Jahn of the Princeton Engineering Anomalies Research group at Princeton University. For more than 20 years he has been carrying out research into 鈥渕icro psychokinesis鈥�, where subjects attempt to influence the output of a random event generator to produce statistically significant deviations from the expected output.
Jahn鈥檚 analysis of his entire data pool for the past 12 years is difficult to argue with. Subjects seem genuinely to succeed in nudging the random number generator in the desired direction. Jahn concludes that the probability of getting the same results by chance are less than 0.00007 鈥� a hugely significant result. And critics struggle to find holes in the methodology or the data.
But attempts to replicate the results have failed. Physicist Stanley Jeffers of York University in Canada tried to do so using a different type of random event generator and got nothing. Jahn himself, in collaboration with German researchers, has tried and failed to replicate the results.
So where do we go from here? Many parapsychologists believe their discipline as presently practised has reached an impasse. They claim that the failure of replication should be taken as a positive result because it confirms what they knew all along: paranormal phenomena are inherently elusive and you can not expect to pin them down in the lab. And increasingly they are reviving pseudo-scientific concepts to explain their data. As a result, the delicate accord of the past 20 years is feeling the strain, as John McCrone discusses on page 34.
Other observers argue that what has been proved is that the debate over parapsychology is inherently intractable. Gathering yet more data will only lead to more controversy, and so is pointless. But there is something positive to be taken from this conclusion, as Robert Matthews argues on page 38: the past 20 years of 鈥渟cientific parapsychology鈥� have taught us about the limits of science itself.